Home » Tech » IBM Migrating COBOL Code into Java using AI

IBM Migrating COBOL Code into Java using AI

IBM Migrating COBOL Code into Java using AI

Since its creation in 1959, COBOL has remained one of the oldest programming languages in use. In a 2022 survey, researchers discovered that the number of lines of COBOL in use on production systems had nearly quadrupled since 2017, with over 800 billion lines. Despite its longevity, COBOL has a reputation for being difficult to understand and inefficient. Now IBM migrating COBOL code into Java

Migrate to a newer platform and take advantage of IBM’s new Code Assistant for IBM Z. This AI model uses code-generation to translate and IBM migrating COBOL code into Java, making the process of modernizing COBOL applications easier and less costly. However, for large organizations, this can still be a complex and costly proposition due to the scarcity of COBOL experts in the world. For example, the Commonwealth Bank of Australia replaced its core COBOL platform in 2012, which took five years and cost over $700 million.

At IBM’s TechXchange conference in Las Vegas this September, Code Assistant for IBM Z will enter preview. IBM Research chief scientist Ruchir Puri designed the tool to help businesses refactor their mainframe apps while protecting performance and security. Code Assistant for IBM Z is expected to become generally available in Q4 2023.

Code Assistant, powered by CodeNet, IBM’s code-generating model, can understand 80 different programming languages and can be used either locally in an on-premises configuration or as a managed service in the cloud. Puri told TechCrunch in an email interview that IBM has designed a cutting-edge generative AI code model to convert legacy COBOL programs to enterprise Java with a high level of accuracy.

Code Assistant not only transforms code, but also supports the entire application modernization process. It helps developers comprehend, restructure, convert, and validate the converted code for a modern architecture.

Video Credits: IBM/Youtube

Parameters are the parts of a model that learn from historical training data and essentially determine the model’s ability to solve a problem, such as generating text, while “tokens” represent raw text — for example “fan,” “tas” and “tic” for the word “fantastic.” As for context window, it refers to the text that the model examines before generating more text.

“This allows you to retain the capabilities of the COBOL language while still reducing costs and making code easier to maintain — something which other offerings on the market fail to do.”

The system will keep a given sub-service of the application in COBOL if its ‘understand’ and ‘refactor’ capabilities recommend it, and it will transform the other sub-services into Java.

Puri warns against using code created by Code Assistant for IBM Z without having it examined by experienced professionals. “The AI system could potentially be unfamiliar with certain patterns of the enterprise’s COBOL application,” Puri stated. “Therefore, it’s essential to use advanced vulnerability scanners to check the code’s security.” Even with potential risks, IBM views tools like Code Assistant as critical for their future development.

Approximately 84% of IBM’s mainframe customers, predominantly from the financial and government sectors, use COBOL. IBM’s mainframe division remains a major part of its business, yet the company is looking to the lucrative hybrid computing environments that it also provides and manages as a pathway forward. Additionally, IBM is aiming to rival apps such as GitHub Copilot and Amazon CodeWhisperer with their own AI-powered code-generating tools.

More News: Related to AI

In May, IBM unveiled fm.model.code within its Watsonx AI service, which powers Watson Code Assistant. This allows developers to use plain English prompts to generate code across programs such as Red Hat’s Ansible Lightspeed.